A transfer function-based digital twin modeling method and system for radio frequency devices

Abstract

The application discloses a kind of radio frequency device digital twin modeling methods based on transfer function. Cubic spline interpolation algorithm is used to obtain the characteristic curve of radio frequency device with continuous change of working frequency of radio frequency signal, and AM-AM curve and AM-PM curve with continuous change of input power of radio frequency signal at each discrete frequency point. Linear transfer function and nonlinear transfer function of radio frequency device are constructed according to the above curves. According to the input signal power of radio frequency device, the working area of radio frequency device is judged, and different transfer functions or fixed parameters are called to obtain intermediate signals. Complex noise is generated, and the same processing method as the input signal of radio frequency device is used for complex noise. Finally, the output signal is obtained. The application realizes the coupling correlation of noise figure and gain compression of radio frequency device.

Description

Technical Field

[0001] This invention relates to a digital twin modeling method for radio frequency devices. Background Technology

[0002] Digital twin technology refers to the construction of a digital simulation model that closely matches the characteristics of a real physical device and channel link, and can be mapped in real time. This model not only includes the static and dynamic transmission characteristics of the device, such as amplitude-frequency, phase-frequency, gain, nonlinearity, and noise, but also reproduces the actual transmission behavior of the device under different operating states, such as the linear region, compression region, and saturation region. It comprehensively reflects the real effects of channel and device noise, amplitude-phase distortion, power-related characteristics, etc., forming a digital twin that corresponds one-to-one with the physical entity and can equivalently replace the physical object for system simulation and verification.

[0003] Radio frequency (RF) devices refer to electronic components or modules capable of processing, converting, amplifying, filtering, or switching radio frequency signals (frequency range between 3kHz and 300GHz). The core of constructing a digital twin of an RF device is building its transfer function. The transfer function of an RF device is a mathematical function that describes the relationship between the input and output signals under different frequencies, temperatures, and voltages. It can accurately reveal the frequency characteristics, gain, phase changes, and multi-characteristic coupling relationships of the RF device.

[0004] In the field of radio frequency (RF) technology, SNP files (TouchStone format) record the S-parameters (scattering parameters), noise figure, power compression characteristics, etc., of RF devices at different frequency points. These are typically discrete data and serve as the foundation for constructing transfer functions. Existing digital twin modeling methods for RF devices based on SNP files can be broadly categorized into the following three types.

[0005] The first type is the direct lookup table method. For parameters at frequency points already recorded in the SNP file, the table is directly looked up and called. For parameters at frequency points not recorded in the SNP file, the corresponding parameter of the closest recorded frequency point is used as an approximation. RF devices constructed using this method have extremely low simulation accuracy and completely fail to meet the core requirement of high-precision simulation in digital twins.

[0006] The second type is the simplified linear interpolation method. This method calculates the S-parameters for frequencies not recorded in the SNP file using linear interpolation, based on the S-parameters of the frequencies already recorded in the SNP file. The curve generated by this method through linear interpolation is a piecewise polygonal line with sharp angles and discontinuous first derivatives. It cannot reflect the physical continuity of the characteristic parameters of RF devices as a function of frequency (such as smooth phase transitions and asymptotic changes in noise figure), thus limiting the reliability of the simulation results.

[0007] The third category consists of commercial simulation software such as ADS and HFSS. These software programs offer high simulation accuracy within a single software environment, but their internal calculation logic and parameter interfaces are not publicly accessible. They cannot directly interface with user-defined digital twin links and suffer from inherent "encapsulation and closure" defects.

[0008] The existing SNP file-based digital twin modeling methods for RF devices generally suffer from the following three problems, which restrict the accuracy and practicality of RF device digital twins.

[0009] First, the coupling relationships between various characteristic parameters of RF devices are lacking. The linear characteristics (S-parameters), noise characteristics (noise figure), and nonlinear characteristics (input power at the 1dB compression point, etc.) of RF devices are interrelated and tightly coupled. For example, when an amplifier operates in the power compression region, its noise figure deteriorates as the gain compression increases. Existing technologies treat these parameters in isolation, failing to reflect the coupling relationships between them, leading to large discrepancies between simulation results and actual conditions under complex operating conditions.

[0010] Second, there is a lack of transfer functions that can characterize the continuous characteristics across the entire frequency band. The SNP file of an RF device records the characteristic parameters of the RF device at multiple discrete frequency points. Existing technologies have not constructed high-precision, continuous, and smooth transfer functions based on these discrete data. Therefore, it is impossible to accurately characterize the full-band characteristics of RF devices—especially in frequency bands with high parameter change rates (such as near the resonant point), and the simulation accuracy cannot meet engineering requirements.

[0011] Third, there are obstacles to digital link interaction. Digital twins need to work collaboratively with other modules through standardized signal links (including carrier frequency, IQ signals, etc.). Existing digital twin modeling of RF devices usually only outputs a single parameter (such as gain value), which is not adapted to the standardized signal links of digital twins. Additional format conversion is required, which not only introduces conversion errors but also reduces the real-time performance of simulations in dynamic scenarios. Summary of the Invention

[0012] The technical problem to be solved by this invention is: how to improve the design of transfer function in digital twin modeling of radio frequency devices, enhance the full-band modeling accuracy of radio frequency devices, fully characterize the coupling relationship between parameters of radio frequency devices, and solve digital link adaptation obstacles.

[0013] To address the aforementioned technical problems, this invention proposes a digital twin modeling method for radio frequency (RF) devices based on transfer functions, comprising the following steps: Step S1: Read the forward transmission gain, forward transmission phase, 1dB compression point input power, saturated input power, and noise figure of the RF device at various discrete frequency points. Use a cubic spline interpolation algorithm to interpolate these RF device characteristic parameters to obtain the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturated input power curve, and noise figure curve of the RF device that continuously vary with the operating frequency of the RF signal. Step S2: At each discrete frequency point, input IQ signals of different powers to the RF device. Obtain AM-AM data and AM-PM data for each discrete input power point through testing. Use a cubic spline interpolation algorithm to interpolate the AM-AM data and AM-PM data for each discrete input power point to obtain the AM-AM curve and AM-PM curve that continuously vary with the input power of the RF signal at each discrete frequency point. Steps S1 and S2 can be performed sequentially or simultaneously, or arbitrarily in the order of steps S1 and S2. Step S3: Based on the AM-AM and AM-PM curves that continuously vary with the input power of the RF signal at each discrete frequency point, the saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, and saturation phase distortion curve of the RF device that continuously varies with the operating frequency of the RF signal are obtained through fitting and cubic spline interpolation algorithms. Step S4: Based on the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturation input power curve, noise figure curve, saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, saturation phase distortion curve of the RF device that continuously varies with the operating frequency of the RF signal, as well as the AM-AM and AM-PM curves that continuously vary with the input power of the RF signal at each discrete frequency point, the linear transfer function and nonlinear transfer function of the RF device are constructed. The linear transfer function characterizes the amplitude linearity and phase linearity characteristics of the RF device that continuously varies with the operating frequency of the RF signal. The nonlinear transfer function characterizes the amplitude and phase nonlinear characteristics of an RF device as the input power of the RF signal continuously changes. The nonlinear transfer function has different expressions in the linear, compressed, and saturated regions of the RF device. Step S5: Determine whether the RF device is in the linear, compressed, or saturated region based on the input signal power. When the RF device is in the linear region, only the linear transfer function is used to calculate the intermediate signal. When the RF device is in the compressed region, only the nonlinear transfer function is used to calculate the intermediate signal. When the RF device is in the saturated region, the fixed parameters for the saturated region are used to obtain the intermediate signal.Step S6: Generate complex noise; depending on whether the RF device is in the linear region, compression region, or saturation region, process the complex noise in the same way as the input signal of the RF device to obtain complex noise processed by the transmission characteristics of the RF device; superimpose the complex noise processed by the transmission characteristics of the RF device onto the intermediate signal to obtain the output signal.

[0014] Furthermore, in step S3, P is extracted from the AM-AM curve of a discrete frequency point that continuously varies with the input power of the radio frequency signal. in ≥P insat (f) data, where P in P represents the input signal power. insat (f) represents the saturation input power of the RF device at that discrete frequency point; for P in ≥P insat (f) The saturation gain of the RF device at that discrete frequency point is obtained by fitting the data in segment (f); the saturation gain G of the RF device at each discrete frequency point is obtained in the same way. sat (f) The saturation gain curve of the RF device as the operating frequency of the RF signal continuously changes is obtained using a cubic spline interpolation algorithm. In step S3, P is extracted from the AM-AM curve of the RF signal input power continuously changing at a certain discrete frequency point. 1dB (f) < P in <P insat (f) data, where P 1dB (f) represents the 1dB compression point input power of the RF device at this discrete frequency; P 1dB (f) < P in <P insat The formula corresponding to the data in segment (f) , where |S 21 (f) represents the forward transmission gain of the RF device at this discrete frequency point; the compression coefficient of the RF device at this discrete frequency point is obtained by fitting using the least squares method; the compression coefficient α(f) of the RF device at each discrete frequency point is obtained in the same way; the compression coefficient curve of the RF device as the operating frequency of the RF signal continuously changes is obtained by using the cubic spline interpolation algorithm. In step S3, P is extracted from the AM-PM curve of the RF signal input power continuously changing at a certain discrete frequency point. 1dB (f) < P in <P insat (f) data, P 1dB (f) < P in <P insat The formula corresponding to the data in segment (f) The first and second phase coefficients of the RF device at the discrete frequency point are obtained by fitting using the least squares method; the first phase coefficient β(f) and second phase coefficient γ(f) of the RF device at each discrete frequency point are obtained in the same way; the first and second phase coefficient curves of the RF device as the operating frequency of the RF signal continuously changes are obtained by using a cubic spline interpolation algorithm. In step S3, the saturated input power P of the RF device at the discrete frequency point is extracted from the AM-PM curve of the RF signal input power continuously changing at a certain discrete frequency point. insat (f) The corresponding phase offset is called the saturation phase distortion value; the saturation phase distortion value of the RF device at each discrete frequency point is obtained in the same way; the saturation phase distortion curve of the RF device when the operating frequency of the RF signal changes continuously is obtained by using the cubic spline interpolation algorithm.

[0015] Furthermore, the feature is that, in step S4, constructing the linear transfer function of the RF device includes the following steps A1 to A3. Step A1: Convert the forward transmission gain curve of the RF device into a complex function of forward transmission gain |H linear (f)|, converting the forward propagation phase curve of the RF device into a complex function of the forward propagation phase ∠H linear (f); ; where, |S 21 (f)| is the forward transmission gain of an RF device corresponding to a certain operating frequency f, obtained from the forward transmission gain curve; ; where ∠S 21 (f) is the forward transmission phase of the RF device corresponding to a certain operating frequency f, obtained from the forward transmission phase curve. Step A2: Calculate the complex function ∠H for the forward transmission phase of the RF device. linear (f) Differentiate to obtain the continuous group delay function τ(f); Step A3: Complex function of the forward transmission gain of the synthesized RF devices |H linear (f)|, Complex function of the forward propagation phase of the radio frequency device ∠H linear (f) The continuous group delay function τ(f) yields the complex form of the linear transfer function H of the RF device. linear (f); Among them, the amplitude term |H linear (f)| Used to apply a frequency-dependent linear gain to the input signal, phase term ∠H linear (f) is used to apply a frequency-dependent linear phase shift to the input signal, and the time delay term -2πfτ(f) is used to correct the time shift of the input signal.

[0016] Further, in step S4, constructing the nonlinear transfer function of the RF device includes the following steps B1 to B3. Step B1: Based on the forward transmission gain curve, 1dB compression point input power curve, saturated input power curve, saturated gain curve, and compression coefficient curve of the RF device, construct the piecewise gain correction function G. nonlinear (f,P in ). Where f is the operating frequency of the radio frequency signal, and P in It is the input signal power, P 1dB (f) is the 1dB compression point input power obtained from the 1dB compression point input power curve for a certain operating frequency f, P. insat (f) is the saturated input power corresponding to a certain operating frequency f obtained from the saturated input power curve, |S 21 (f)| is the forward transmission gain of the RF device corresponding to a certain operating frequency f, obtained from the forward transmission gain curve; α(f) is the compression coefficient corresponding to a certain operating frequency f, obtained from the compression coefficient curve; G sat (f) is to obtain the saturation gain corresponding to a certain operating frequency f from the saturation gain curve. Step B2: Based on the 1dB compression point input power curve, saturation phase distortion curve, first phase coefficient curve, and second phase coefficient curve of the RF device, construct the phase distortion function Δφ(f,P) in ). Where f is the operating frequency of the radio frequency signal, and P in It is the input signal power, P 1dB (f) is the 1dB compression point input power obtained from the 1dB compression point input power curve, β(f) is the first phase coefficient obtained from the first phase coefficient curve, γ(f) is the second phase coefficient obtained from the second phase coefficient curve, and Δφ sat (f) The saturated phase distortion value corresponding to a certain operating frequency f is obtained by querying the saturated phase distortion curve. Steps B1 and B2 can be performed simultaneously or arbitrarily in sequence. Step B3: Synthesize the piecewise gain correction function G. nonlinear (f,P in ), Phase distortion function Δφ(f,P in The complex form of the nonlinear transfer function H of the radio frequency device is obtained. nonlinear (f,P in ). Where j is the imaginary unit; the real part represents the amplitude nonlinear distortion of the RF device, and the imaginary part represents the phase nonlinear distortion of the RF device; the nonlinear transfer function has three rows, corresponding to the linear region, compression region, and saturation region of the RF device, respectively; in the compression region of the RF device, the phase distortion of the imaginary part is proportional to the "difference between the input power and the input power at the 1dB compression point and the second phase coefficient", allowing the phase distortion to be dynamically adjusted with the input signal power.

[0017] Furthermore, in step S5, the carrier frequency f and instantaneous power P of the radio frequency device are calculated from the input signal. in A 3dB bandwidth B is used. When the input signal is a single carrier, the carrier frequency is extracted using Fast Fourier Transform (FFT). When the input signal is multi-carrier, the subcarrier frequency set is extracted using Discrete Fourier Transform (DFT). Based on the carrier frequency or subcarrier frequency of the input signal, the 1dB compression point input power corresponding to the carrier frequency or subcarrier frequency is obtained from the 1dB compression point input power curve as the linear region threshold P1, and the saturation input power corresponding to the carrier frequency or subcarrier frequency is obtained from the saturation input power curve as the saturation region threshold P2. The input signal power P is compared. in Using the linear region threshold P1 and the saturation region threshold P2, determine which operating region the RF device is in; if P in ≤P1 indicates that the RF device is in the linear region; if P1 < P in <P2 indicates that the RF device is in the compressed region; if P in ≥P2 indicates that the radio frequency device is in the saturation region.

[0018] Furthermore, in step S5, when the RF device is in the linear region, if the input signal is a single-carrier signal, first perform an FFT on the input signal to obtain the frequency domain component, then call the linear transfer function, and then perform an inverse fast Fourier transform (IFFT) to convert it back to the time domain to obtain the intermediate signal; if the input signal is a multi-carrier signal, first perform an FFT on the input signal to obtain the frequency domain component, then call the linear transfer function for each subcarrier, and finally perform an IFFT to convert it back to the time domain to obtain the intermediate signal.

[0019] Furthermore, in step S5, when the radio frequency device is in the compression region, the input signal is first subjected to FFT to obtain the frequency domain component, then the nonlinear transfer function is called, and finally IFFT is performed to convert it back to the time domain to obtain the intermediate signal.

[0020] Furthermore, in step S5, when the radio frequency device is in the saturation region, the fixed parameters of the saturation region are called to output an intermediate signal.

[0021] Further, step S6 specifically includes the following steps S61 to S65. Step S61: Calculate the noise power spectral density function N0 based on the carrier frequency f and 3dB bandwidth B of the input signal of the RF device; convert the noise power spectral density function N0 into a linear power P. n Step S62: Distribute the complex noise power to the I and Q channels, and calculate the power P of each channel. nI and P nQ Variance of each path Step S63: Generate two independent zero-mean values ​​with powers P and P respectively. nI and P nQ , and variance are respectively and A Gaussian white noise sequence, i.e., I-channel in-phase noise n I Orthogonal noise of Q-path n Q The combination forms time-domain complex noise n I +jn Q Step S64: If the input signal of the RF device is a single-carrier signal and the RF device is in the linear region, only the linear transfer function is applied to the time-domain complex noise to obtain complex noise processed by the RF device's transmission characteristics. If the input signal of the RF device is a single-carrier signal and the RF device is in the compression region, only the nonlinear transfer function is applied to the time-domain complex noise to obtain complex noise processed by the RF device's transmission characteristics. If the input signal of the RF device is a single-carrier signal and the RF device is in the saturation region, fixed parameters are used for output in the saturation region to obtain complex noise processed by the RF device's transmission characteristics. If the input signal of the RF device is a multi-carrier signal and the RF device is in the linear region, the time-domain complex noise is first converted to frequency-domain noise using FFT, only the linear transfer function is applied to the frequency-domain noise, and then the processed frequency-domain noise is converted back to the time domain using IFFT to obtain complex noise processed by the RF device's transmission characteristics. If the input signal of the RF device is a multi-carrier signal and the RF device is in the compression region, the time-domain complex noise is first converted to frequency-domain noise using FFT. Only the nonlinear transfer function is applied to the frequency-domain noise. Then, the processed frequency-domain noise is converted back to the time domain using IFFT, resulting in complex noise processed by the RF device's transmission characteristics. If the input signal of the RF device is a multi-carrier signal and the RF device is in the saturation region, the time-domain complex noise is first converted to frequency-domain noise using FFT. The frequency-domain noise is output using fixed parameters in the saturation region. Then, the output frequency-domain noise is converted back to the time domain using IFFT, resulting in complex noise processed by the RF device's transmission characteristics. Step S65: The complex noise processed by the RF device's transmission characteristics is superimposed onto the intermediate signal to obtain the output signal. The output signal is encapsulated using the standardized signal link format of the digital prototype.

[0022] This invention also proposes a digital twin modeling system for radio frequency (RF) devices based on transfer functions, including a continuous frequency curve generation module one, a continuous power curve generation module two, a transfer function construction module, a signal transmission calculation module, and a noise transmission calculation module. The continuous frequency curve generation module one is used to read the forward transmission gain, forward transmission phase, 1dB compression point input power, saturated input power, and noise figure of the RF device at various discrete frequency points. A cubic spline interpolation algorithm is used to interpolate these RF device characteristic parameters to obtain the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturated input power curve, and noise figure curve of the RF device that continuously change with the operating frequency of the RF signal. The continuous power curve generation module is used to input IQ signals of different powers to the RF device at each discrete frequency point. AM-AM data and AM-PM data are obtained through testing at each discrete input power point. A cubic spline interpolation algorithm is used to interpolate the AM-AM data and AM-PM data at each discrete input power point to obtain the AM-AM curve and AM-PM curve that continuously change with the input power of the RF signal at each discrete frequency point. The continuous frequency curve generation module 2 is used to obtain the saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, and saturation phase distortion curve of the radio frequency device that continuously varies with the operating frequency of the radio frequency signal by fitting and cubic spline interpolation algorithm based on the AM-AM curve and AM-PM curve that continuously vary with the input power of the radio frequency signal at each discrete frequency point. The transfer function construction module is used to construct the linear and nonlinear transfer functions of an RF device based on the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturated input power curve, noise figure curve, saturated gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, saturated phase distortion curve, and AM-AM and AM-PM curves of the RF device continuously varying with the input power of the RF signal at each discrete frequency point. The linear transfer function characterizes the amplitude linearity and phase linearity of the RF device continuously varying with the operating frequency of the RF signal; the nonlinear transfer function characterizes the amplitude nonlinearity and phase nonlinearity of the RF device continuously varying with the input power of the RF signal; the nonlinear transfer function has different expressions in the linear region, compression region, and saturation region of the RF device. The signal transmission calculation module is used to determine whether the RF device is in the linear region, compression region, or saturation region based on the input signal power of the RF device. When the RF device is in the linear region, it uses only the linear transfer function to calculate the intermediate signal. When the RF device is in the compression region, it uses only the nonlinear transfer function to calculate the intermediate signal. When the RF device is in the saturation region, it uses the fixed parameters of the saturation region to obtain the intermediate signal.The noise transmission calculation module is used to generate complex noise; depending on whether the RF device is in the linear region, compression region, or saturation region, the complex noise is processed in the same way as the input signal of the RF device to obtain complex noise processed by the transmission characteristics of the RF device; the complex noise processed by the transmission characteristics of the RF device is superimposed on the intermediate signal to obtain the output signal.

[0023] The technical effect achieved by this invention is that, in the nonlinear transfer function of the radio frequency device, a fitting formula including the "difference between the input power and the input power at the 1dB compression point" is used for the compression region of the radio frequency device, thereby realizing the coupling relationship between the noise figure and gain compression of the radio frequency device. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the digital twin modeling method for radio frequency devices based on transfer functions proposed in this invention.

[0025] Figure 2 yes Figure 1 The diagram below illustrates the specific process of constructing the linear transfer function of the RF device in step S4.

[0026] Figure 3 yes Figure 1 The diagram below illustrates the specific process of constructing the nonlinear transfer function of the radio frequency device in step S4.

[0027] Figure 4 yes Figure 1 The detailed flowchart of step S6 is shown below.

[0028] Figure 5 This is a schematic diagram of the structure of the digital twin modeling system for radio frequency devices based on transfer function proposed in this invention.

[0029] The following are the annotations in the figure: 1. Continuous frequency curve generation module 1; 2. Continuous power curve generation module 2; 3. Continuous frequency curve generation module 3; 4. Transfer function construction module; 5. Signal transmission calculation module; 6. Noise transmission calculation module. Detailed Implementation

[0030] Please see Figure 1 The digital twin modeling method for radio frequency devices based on transfer function proposed in this invention includes the following steps.

[0031] Step S1: Read the forward propagation gain |S of the RF device at each discrete frequency point (the operating frequency of the RF signal). 21 |、Forward transmission phase ∠S 21 1dB compression point input power P 1dB Saturated input power P insatNoise figure NF. The cubic spline interpolation algorithm is used to interpolate the above RF device characteristic parameters to obtain the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturation input power curve, and noise figure curve of the RF device that change continuously with the operating frequency of the RF signal.

[0032] Step S2: At each discrete frequency point, input IQ signals of different powers to the RF device. The IQ signals are divided into an I-channel in-phase signal and a Q-channel quadrature signal. AM-AM (amplitude modulation - amplitude modulation) data and AM-PM (amplitude-to-phase) data for each discrete input power point are obtained through testing. Cubic spline interpolation is used to interpolate the AM-AM and AM-PM data for each discrete input power point, respectively, to obtain the AM-AM curve and AM-PM curve for each discrete frequency point that continuously change with the input power of the RF signal.

[0033] The order of steps S1 and S2 is not strictly limited; they can be performed simultaneously or either can be performed first or last.

[0034] Step S3: Based on the AM-AM curve and AM-PM curve that continuously change with the input power of the RF signal at each discrete frequency point, the saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, and saturation phase distortion curve of the RF device that continuously change with the operating frequency of the RF signal are obtained by fitting and cubic spline interpolation algorithm.

[0035] Step S4: Based on the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturation input power curve, noise figure curve, saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, saturation phase distortion curve, and AM-AM and AM-PM curves of the RF device continuously varying with the RF signal's operating frequency, the linear and nonlinear transfer functions of the RF device are constructed. The linear transfer function characterizes the amplitude linearity and phase linearity of the RF device as the RF signal's operating frequency continuously changes. The nonlinear transfer function characterizes the amplitude nonlinearity and phase nonlinearity of the RF device as the RF signal's input power continuously changes. The linear and nonlinear transfer functions of this invention support both time-domain and frequency-domain calculations. The nonlinear transfer function has different expressions in the linear, compression, and saturation regions of the RF device.

[0036] Step S5: Determine whether the RF device is in the linear region, compression region, or saturation region based on the input signal power. If the RF device is in the linear region, calculate the intermediate signal using only the linear transfer function. If the RF device is in the compression region, calculate the intermediate signal using only the nonlinear transfer function. If the RF device is in the saturation region, use the fixed parameters for the saturation region to obtain the intermediate signal.

[0037] Step S6: Generate complex noise n I +jn Q Where n I Indicates the in-phase noise of the I-channel, n Q This represents the Q-path quadrature noise. Depending on whether the RF device is in the linear, compressed, or saturated region, the complex noise is processed in the same way as the input signal of the RF device, resulting in complex noise processed by the RF device's transmission characteristics. This processed complex noise is then superimposed onto the intermediate signal to obtain the output signal.

[0038] In step S1, the SNP file of the RF device stores the characteristic parameters of the RF device in discrete data form. By reading the SNP file of the RF device, the amplitude and phase characteristics of the S-parameters of the RF device at each discrete frequency point and the input power P at the 1dB compression point can be obtained. 1dB Saturated input power P insat Noise figure NF. The amplitude and phase characteristics of the S-parameters specifically include S... 11 S 21 S 12 S 22 The amplitude and phase characteristics. S 11 S reflects the input matching characteristics 22 S reflects the output matching characteristics 21 S reflects the forward transmission gain and phase characteristics. 12 This reflects the reverse isolation characteristics. Among them, S 21 The amplitude and phase characteristics of the parameters are essential information that must be read in this invention. 11 Parameters, S 12 Parameters, S 22 The amplitude and phase characteristics of the parameters are optional. 21 The amplitude characteristic of a parameter refers to the forward transmission gain of an RF device, which is the ratio of output signal power to input signal power. 21 The phase characteristic of a parameter refers to the forward propagation phase of an RF device, that is, the change in phase between the output signal and the input signal. The input power P at the 1dB compression point. 1dB This refers to the input power value that occurs when the actual gain of the RF device decreases by 1 dB compared to its small-signal linear gain (theoretical constant gain) as the input signal power increases. Saturated input power Pinsat The noise figure (NF) refers to the input power required for an RF device to transition from the compressed region to the saturated region as the input signal power increases; it characterizes the boundary between the compressed and saturated regions of an RF device. The noise figure (NF) is the decibel representation of the noise factor (F). NF = 10log 10 (F), the unit is dB. The noise factor F refers to the signal-to-noise ratio (SNR) of the input signal of an RF device. in With output signal-to-noise ratio (SNR) out The ratio, F = SNR in ÷SNR out .

[0039] In step S1, the cubic spline interpolation algorithm constructs piecewise cubic polynomials to ensure that the interpolation curves are continuous and smooth when the operating frequency of the RF signal changes continuously across the entire frequency band, and that both the first and second derivatives are continuous, perfectly matching the forward transmission gain |S| of the RF device. 21 |、Forward transmission phase ∠S 21 1dB compression point input power P 1dB Saturated input power P insat The noise figure (NF) exhibits a physical smoothness characteristic that continuously varies with the operating frequency of the RF signal. The horizontal axis of the forward transmission gain curve represents the operating frequency of the RF signal, and the vertical axis represents the forward transmission gain of the RF device. The horizontal axis of the forward transmission phase curve represents the operating frequency of the RF signal, and the vertical axis represents the forward transmission phase of the RF device. The horizontal axis of the 1dB compression point input power curve represents the operating frequency of the RF signal, and the vertical axis represents the 1dB compression point input power of the RF device. The horizontal axis of the saturated input power curve represents the operating frequency of the RF signal, and the vertical axis represents the saturated input power of the RF device. The horizontal axis of the noise figure curve represents the operating frequency of the RF signal, and the vertical axis represents the noise figure of the RF device.

[0040] In step S2, AM-AM data is used to describe the amplitude nonlinear distortion of the RF device, reflecting the relationship between the instantaneous output gain of the RF device and the input power. AM-PM data is used to describe the phase nonlinear distortion of the RF device, reflecting the relationship between the phase offset of the output signal of the RF device and the input power.

[0041] In step S2, the cubic spline interpolation algorithm constructs a piecewise cubic polynomial to ensure that the interpolation curves are continuous and smooth when the input power of the RF signal changes continuously, and that both the first and second derivatives are continuous. This perfectly matches the physical smoothness of the AM-AM and AM-PM characteristics of the RF device as the input power of the RF signal changes continuously. Each discrete frequency point of the RF signal has one AM-AM curve and one AM-PM curve, and multiple (e.g., N) discrete frequency points of the RF signal have N AM-AM curves and N AM-PM curves. The horizontal axis of the AM-AM curve represents the input signal power, and the vertical axis represents the instantaneous output gain of the RF device. A typical AM-AM curve is divided into three operating regions: the linear region, the compression region, and the saturation region. The input power P at the 1dB compression point... sat It is a key indicator for measuring the compression range. Saturated input power P insat It is a key indicator in the saturation region. The horizontal axis of the AM-PM curve represents the input signal power, and the vertical axis represents the phase shift of the output signal.

[0042] In step S3, P is extracted from the AM-AM curve at a certain discrete frequency point. in ≥P insat The data in section (f) shows the input signal power on the horizontal axis and the instantaneous output gain of the RF device on the vertical axis. Where P... in P represents the input signal power. insat (f) represents the saturation input power of the RF device at that discrete frequency point. For P in ≥P insat The saturation gain of the RF device at that discrete frequency point is obtained by fitting the data in segment (f). The saturation gain G of the RF device at each discrete frequency point is obtained in the same way. sat (f). The saturation gain curve of the RF device as the operating frequency of the RF signal changes continuously is obtained by using the cubic spline interpolation algorithm.

[0043] In step S3, P is extracted from the AM-AM curve at a certain discrete frequency point. 1dB (f) < P in <P insat The data in section (f) shows the input signal power on the horizontal axis and the instantaneous output gain of the RF device on the vertical axis. Where P... 1dB (f) represents the 1dB compression point input power of the RF device at this discrete frequency. P 1dB (f) < P in <P insat The formula corresponding to the data in segment (f) , where |S 21(f) represents the forward transmission gain of the RF device at this discrete frequency point. The compression coefficient of the RF device at this discrete frequency point is obtained by fitting using the least squares method. The compression coefficient α(f) of the RF device at each discrete frequency point is obtained in the same way. The compression coefficient curve of the RF device as the operating frequency of the RF signal changes continuously is obtained by using a cubic spline interpolation algorithm.

[0044] In step S3, P is extracted from the AM-PM curve at a certain discrete frequency point. 1dB (f) < P in <P insat The data in segment (f) shows the input signal power on the horizontal axis and the phase offset of the output signal on the vertical axis. P... 1dB (f) < P in <P insat The formula corresponding to the data in segment (f) The first and second phase coefficients of the RF device at the discrete frequency points are obtained by fitting using the least squares method. The first phase coefficient β(f) and second phase coefficient γ(f) of the RF device at each discrete frequency point are obtained in the same way. The first and second phase coefficient curves of the RF device as the operating frequency of the RF signal continuously changes are obtained by using a cubic spline interpolation algorithm.

[0045] In step S3, the saturated input power P of the radio frequency device at a discrete frequency point is extracted from the AM-PM curve at that discrete frequency point. insat (f) The corresponding phase offset is called the saturation phase distortion value. The saturation phase distortion values ​​of the RF device at each discrete frequency point are obtained using the same method. A cubic spline interpolation algorithm is used to obtain the saturation phase distortion curve of the RF device as the operating frequency of the RF signal continuously changes.

[0046] In step S3, the cubic spline interpolation algorithm constructs piecewise cubic polynomials to ensure that the interpolation curves are continuous and smooth when the operating frequency of the RF signal changes continuously across the entire frequency band, and that both the first and second derivatives are continuous, perfectly matching the saturation gain G of the RF device. sat(f) The compression coefficient α(f), the first phase coefficient β(f), the second phase coefficient γ(f), and the physical smoothness characteristic of the saturation phase distortion value continuously changing with the operating frequency of the RF signal. The horizontal axis of the saturation gain curve is the operating frequency of the RF signal, and the vertical axis is the saturation gain of the RF device. The horizontal axis of the compression coefficient curve is the operating frequency of the RF signal, and the vertical axis is the compression coefficient of the RF device. The horizontal axis of the first phase coefficient curve is the operating frequency of the RF signal, and the vertical axis is the first phase coefficient of the RF device. The horizontal axis of the second phase coefficient curve is the operating frequency of the RF signal, and the vertical axis is the second phase coefficient of the RF device. The horizontal axis of the saturation phase distortion curve is the operating frequency of the RF signal, and the vertical axis is the saturation phase distortion value of the RF device.

[0047] Please see Figure 2 The construction of the linear transfer function of the radio frequency device in step S4 includes the following steps A1 to A3.

[0048] Step A1: Convert the forward transmission gain curve of the RF device into a complex function of forward transmission gain |H linear (f)|, converting the forward propagation phase curve of the RF device into a complex function of the forward propagation phase ∠H linear (f). Among them, |S 21 (f)| is the forward transmission gain of an RF device at a certain operating frequency f obtained from the forward transmission gain curve, and the unit is dB (decibels). Among them, ∠S 21 (f) is the forward transmission phase of the RF device corresponding to a certain operating frequency f, obtained from the forward transmission phase curve, and the unit is degrees.

[0049] Step A2: Complex function of the forward propagation phase ∠H for the RF device linear (f) Differentiating, we obtain the continuous group delay function τ(f). The group delay function τ(f) is defined as the forward propagation phase ∠H of the RF device. linear (f) The negative derivative with respect to the operating frequency f of the radio frequency signal. .

[0050] Step A3: Complex function of forward transmission gain of integrated RF devices |H linear (f)|, Complex function of the forward propagation phase of the radio frequency device ∠H linear (f) The continuous group delay function τ(f) yields the complex form of the linear transfer function H of the RF device. linear (f). Among them, the amplitude term |H linear (f)| Used to apply a frequency-dependent linear gain to the input signal, phase term ∠H linear(f) is used to apply a frequency-dependent linear phase shift to the input signal, and the time delay term -2πfτ(f) is used to correct the time shift of the input signal.

[0051] Please see Figure 3 The construction of the nonlinear transfer function of the radio frequency device in step S4 includes the following steps B1 to B3.

[0052] Step B1: Based on the forward transmission gain curve, 1dB compression point input power curve, saturation input power curve, saturation gain curve, and compression coefficient curve of the RF device, construct the piecewise gain correction function G. nonlinear (f,P in ).

[0053] Where f is the operating frequency of the radio frequency signal, and P... in It is the input signal power. P 1dB (f) is the input power at the 1dB compression point obtained from the input power curve at the 1dB compression point, corresponding to a certain operating frequency f. insat (f) is the saturated input power corresponding to a certain operating frequency f obtained by querying the saturated input power curve. |S 21 (f)| is the forward transmission gain of the RF device at a given operating frequency f, obtained from the forward transmission gain curve, in dB. α(f) is the compression factor at a given operating frequency f, obtained from the compression factor curve. G sat (f) is the saturation gain corresponding to a certain operating frequency f obtained from the saturation gain curve.

[0054] Step B2: Based on the 1dB compression point input power curve, saturation phase distortion curve, first phase coefficient curve, and second phase coefficient curve of the RF device, construct the phase distortion function Δφ(f,P) in ).

[0055] Where f is the operating frequency of the radio frequency signal, and P... in It is the input signal power. P 1dB (f) is the 1dB compression point input power obtained from the 1dB compression point input power curve for a given operating frequency f. β(f) is the first phase coefficient obtained from the first phase coefficient curve for a given operating frequency f. γ(f) is the second phase coefficient obtained from the second phase coefficient curve for a given operating frequency f. Δφ sat (f) is the saturation phase distortion value corresponding to a certain operating frequency f obtained by querying the saturation phase distortion curve.

[0056] The order of steps B1 and B2 is not strictly limited; they can be performed simultaneously or either one can be performed first.

[0057] Step B3: Synthesize the piecewise gain correction function G nonlinear (f,P in ), Phase distortion function Δφ(f,P in The complex form of the nonlinear transfer function H of the radio frequency device is obtained. nonlinear (f,P in The nonlinear transfer function is used to characterize the amplitude nonlinear distortion and phase nonlinear distortion of radio frequency devices.

[0058] Where j is the imaginary unit. The real part represents the amplitude nonlinear distortion of the RF device, and the imaginary part represents the phase nonlinear distortion of the RF device. The nonlinear transfer function has three expressions (three lines), corresponding to the linear, compressed, and saturated regions of the RF device, respectively. In practical applications, the corresponding expressions of the nonlinear transfer function are not used in the linear and saturated regions; only the compressed region uses the corresponding expressions of the nonlinear transfer function. The e in the first line on the right side of the equals sign... j0 This indicates that the phase distortion of the RF device is zero in the linear region. By setting the phase distortion in the linear region to zero, the nonlinear transfer function in the linear region retains only the real-valued gain, accurately reflecting the characteristic that there is no additional phase change in the linear region.

[0059] Traditional digital twin modeling techniques for radio frequency (RF) devices generally suffer from the following shortcomings in handling phase distortion: First, most employ a single modeling approach that does not differentiate between operating regions, failing to distinguish between the linear, compressed, and saturated regions of the RF device, and using a uniform expression to describe the phase distortion. For example, a fixed phase value is superimposed on the linear region of the RF device, even though there is no additional phase shift. Similarly, the phase distortion in the compressed region of the RF device does not change with the input signal power. This contradicts the physical law that actual RF devices have different phase characteristics in different operating regions. Second, even when AM-PM characteristics are considered, the phase coefficients of the fitted AM-PM curve are mostly fixed constants, independent of the carrier frequency. This contradicts the inherent property that the phase characteristics of actual RF devices dynamically change with the operating frequency. Third, there is a correlation between the phase distortion of RF devices and other characteristics. For example, as the gain compression intensifies, the phase distortion changes synchronously. Existing technologies treat phase distortion in isolation, ignoring the "coupling effect between gain compression and phase distortion," leading to significant deviations between the model and the actual characteristics of the RF device.

[0060] This invention makes targeted improvements to the handling of phase distortion when constructing the nonlinear transfer function of radio frequency (RF) devices. First, this invention overcomes the limitations of a unified expression by designing differentiated phase distortion functions that match the actual characteristics of the RF device in its linear, compressed, and saturated regions. In the linear region of the RF device, there is no phase distortion, so the phase distortion is directly set to zero, closely reflecting reality. In the compressed region of the RF device, the phase distortion varies with power; a fitting formula of "the difference between the input power and the input power at the 1dB compression point and the frequency correlation coefficient" is used to dynamically adjust the phase distortion according to the input signal power. In the saturated region of the RF device, the phase characteristics are stable; the phase distortion is fixed to the saturated phase value, matching the characteristics of the RF device in the saturated region. Second, the two phase coefficients β(f) and γ(f) in this invention are not fixed constants, but are obtained based on AM-PM curve fitting and are dynamic parameters that vary with the carrier frequency (the operating frequency of the RF signal). This solves the problem of "phase coefficients being out of sync with frequency" in existing technologies, and better reflects the frequency characteristics of actual RF devices. Third, the phase distortion function constructed in this invention is deeply bound to the "gain compression degree". The phase distortion value of the RF device in the compression region is determined by the "difference between the input power and the input power at the 1dB compression point" (i.e., the gain compression degree), realizing the coupling effect of "deepening gain compression → synchronous change in phase distortion", avoiding the defect of "isolated modeling of phase distortion" in the prior art.

[0061] In step S5, the carrier frequency f and instantaneous power P of the radio frequency device are calculated from the input signal. in The 3dB bandwidth B is defined as follows: When the input signal is a single carrier, the carrier frequency is extracted using a Fast Fourier Transform (FFT). When the input signal is multi-carrier, the subcarrier frequency set is extracted using a Discrete Fourier Transform (DFT). In this invention, the carrier frequency or subcarrier frequency of the input signal of the RF device is the operating frequency of the RF device. The 3dB bandwidth B refers to the frequency range corresponding to the power drop to half of the maximum value (i.e., -3dB) in the signal power spectral density curve. Based on the carrier frequency (single-carrier scenario) or subcarrier frequency (multi-carrier scenario) of the input signal, the 1dB compression point input power corresponding to the carrier frequency or subcarrier frequency is obtained from the 1dB compression point input power curve as the linear region threshold P1, and the saturated input power corresponding to the carrier frequency or subcarrier frequency is obtained from the saturated input power curve as the saturation region threshold P2.

[0062] In step S5, the input signal power P is compared. in By comparing the linear region threshold P1 and the saturation region threshold P2, we can determine which operating region the RF device is in and use the corresponding processing method to obtain the intermediate signal.

[0063] If P in≤P1 indicates that the output signal has no significant distortion, the RF device is in the linear region, and only linear processing is performed on the input signal. If the input signal is a single-carrier signal, an FFT is first performed on the input signal to obtain the frequency domain component, then a linear transfer function is called to apply linear gain, linear phase shift, and time delay correction, and finally an Inverse Fast Fourier Transform (IFFT) is performed to convert it back to the time domain to obtain the intermediate signal. If the input signal is a multi-carrier signal, an FFT is first performed on the input signal to obtain the frequency domain component, then a linear transfer function is called on each subcarrier to apply linear gain, phase shift, and time delay correction, and finally an IFFT is performed to convert it back to the time domain to obtain the intermediate signal.

[0064] If P1 < P in <P2 indicates that the output signal exhibits gain compression and phase distortion. Since the RF device is in the compression region, nonlinear correction of the input signal is required. Specifically, an FFT is first performed on the input signal to obtain the frequency domain components. Then, a nonlinear transfer function is called to apply amplitude nonlinear distortion (point-by-point gain) and phase nonlinear distortion (point-by-point phase distortion). Finally, an IFFT is performed to transform the signal back to the time domain, yielding the intermediate signal.

[0065] If P in ≥P2 indicates that the output signal is fully saturated. In the saturation region, the gain and phase of the RF device tend to stabilize, and the input signal is processed according to saturation characteristics. Specifically, fixed parameters (fixed gain and fixed phase offset) are used in the saturation region to obtain an intermediate signal. The fixed gain and fixed phase offset no longer change with the input signal power.

[0066] Please see Figure 4 Step S6 specifically includes the following steps S61 to S65.

[0067] Step S61: Calculate the noise power spectral density function N0 based on the carrier frequency f and 3dB bandwidth B of the input signal of the RF device. The unit is dBm (decibels per milliwatt). Here, NF(f) is the noise figure corresponding to the carrier frequency f of the input signal, obtained from the noise figure curve of the RF device. The noise power spectral density function N0 is converted into linear power P. n . The unit is mW (milliwatts).

[0068] Step S62: The power of the complex noise is I-channel in-phase noise n I The power and Q-path quadrature noise n Q The sum of the power of the noise is assumed to be equal for both I and Q channels in engineering practice. Therefore, the complex noise power is allocated to both I and Q channels, and the power of each channel is... The power of Gaussian white noise is equal to its variance. (Assuming the noise mean is 0), therefore the variances of the I and Q noise channels are... , .

[0069] Step S63: Generate two independent values ​​with zero mean and variance respectively. and A Gaussian white noise sequence, i.e., I-channel in-phase noise n I Orthogonal noise of Q-path n Q The combination forms time-domain complex noise n I +jn Q The power of time-domain complex noise is determined by the noise power spectral density function N0.

[0070] Step S64: If the input signal of the RF device is a single-carrier signal and the RF device is in the linear region, only the linear transfer function is called to apply linear gain, linear phase offset and time delay correction to the time-domain complex noise to obtain the complex noise processed by the transmission characteristics of the RF device.

[0071] If the input signal of the RF device is a single-carrier signal and the RF device is in the compression region, only the amplitude nonlinear distortion and phase nonlinear distortion are applied to the time-domain complex noise by calling the nonlinear transfer function, and the complex noise is obtained after being processed by the transmission characteristics of the RF device.

[0072] If the input signal of the RF device is a single-carrier signal and the RF device is in the saturation region, using fixed parameters in the saturation region will produce complex noise processed by the transmission characteristics of the RF device.

[0073] If the input signal of the RF device is a multi-carrier signal and the RF device is in the linear region, the time-domain complex noise is first converted into frequency-domain noise through FFT. The frequency-domain noise is then subjected to linear gain, linear phase shift and time delay correction by only calling the linear transfer function. The processed frequency-domain noise is then converted back to the time domain through IFFT, which is the complex noise processed by the transmission characteristics of the RF device.

[0074] If the input signal of the RF device is a multi-carrier signal and the RF device is in the compression region, the time-domain complex noise is first converted into frequency-domain noise through FFT. The frequency-domain noise is then subjected to amplitude nonlinear distortion and phase nonlinear distortion by calling the nonlinear transfer function. The processed frequency-domain noise is then converted back to the time domain through IFFT, which is the complex noise processed by the transmission characteristics of the RF device.

[0075] If the input signal of the RF device is a multi-carrier signal and the RF device is in the saturation region, the time-domain complex noise is first converted into frequency-domain noise through FFT. The frequency-domain noise is output with fixed parameters in the saturation region. Then, the output frequency-domain noise is converted back to the time domain through IFFT. This is the complex noise processed by the transmission characteristics of the RF device.

[0076] Step S65: The complex noise processed by the transmission characteristics of the radio frequency device is superimposed on the intermediate signal to obtain the output signal. The output signal is encapsulated in the format of the standardized signal link of the digital prototype.

[0077] Traditional digital twin modeling techniques for radio frequency (RF) devices suffer from the following shortcomings in noise generation and superposition methods: First, they fail to differentiate between the operating regions (linear, compressed, and saturated regions) of RF devices, directly superimposing noise with fixed characteristics. This ignores the "modulation" of noise by RF devices in nonlinear regions (such as gain distortion in compressed regions), leading to poor matching between noise and the actual system, and simulation results deviating from reality. Second, they fail to distinguish between single-carrier and multi-carrier input signals, employing a "one-size-fits-all" approach with the same noise superposition logic, which cannot adapt to the time and frequency domain characteristics of different signals. For example, using frequency domain processing for single-carrier signals results in poor fit, while using time domain processing for multi-carrier signals leads to low efficiency. Third, the noise processing does not follow the processing flow of the RF device's input signal, resulting in poor coordination between noise, signal, and device characteristics, failing to accurately reproduce the noise behavior in the system.

[0078] In the digital twin modeling technology for radio frequency (RF) devices of this invention, targeted improvements have been made to the generation and superposition of noise. First, noise is processed accordingly based on the different operating regions of the RF device (e.g., gain compression and phase distortion in the compression region, and fixed amplitude and phase in the saturation region). This perfectly matches the nonlinear characteristics of actual RF devices, avoiding the problem of noise being out of sync with the RF device state, resulting in more realistic simulation and test results. Second, different noise processing procedures are applied depending on whether the input signal of the RF device is a single-carrier or multi-carrier signal. In single-carrier scenarios, noise is processed on a time-domain sampling basis without frequency-domain conversion. In multi-carrier scenarios, subcarrier characteristics are matched in the frequency domain using FFT and IFFT. This adapts to the transmission characteristics of input signals with different carrier forms while also considering processing efficiency. Third, noise processing follows the processing flow of the RF device's input signal. The same processing logic is used for both the input signal and noise depending on the operating region of the RF device before superposition, ensuring complete coordination between the noise and input signal processing logic and the RF device characteristics, more closely resembling the signal-noise interaction process of a real system. Fourth, this invention generates two independent Gaussian noise channels, I and Q, and allocates the noise power of the two channels based on actual power conversion (noise power spectral density function to linear power). This perfectly matches the orthogonal transmission characteristics of I and Q in complex signal systems such as communication and radar, and the statistical characteristics of the noise are highly consistent with the actual complex signal mechanism.

[0079] Please see Figure 5The digital twin modeling system for radio frequency devices based on transfer function proposed in this invention includes a continuous frequency curve generation module 1, a continuous power curve generation module 2, a continuous frequency curve generation module 3, a transfer function construction module 4, a signal transmission calculation module 5, and a noise transmission calculation module 6. Figure 5 The system shown corresponds to Figure 1 The method shown.

[0080] The continuous frequency curve generation module 1 is used to read the positive transmission gain |S| of the radio frequency device at each discrete frequency point. 21 |、Forward transmission phase ∠S 21 1dB compression point input power P 1dB Saturated input power P insat The noise figure NF is obtained by interpolating the above RF device characteristic parameters using a cubic spline interpolation algorithm, resulting in the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturation input power curve, and noise figure curve of the RF device that change continuously with the operating frequency of the RF signal.

[0081] The continuous power curve generation module 2 is used to input IQ signals of different power to the radio frequency device at each discrete frequency point. The AM-AM data and AM-PM data of each discrete input power point are obtained through testing. The cubic spline interpolation algorithm is used to interpolate the AM-AM data and AM-PM data of each discrete input power point to obtain the AM-AM curve and AM-PM curve that change continuously with the input power of the radio frequency signal at each discrete frequency point.

[0082] The continuous frequency curve generation module 3 is used to obtain the saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, and saturation phase distortion curve of the radio frequency device that continuously varies with the operating frequency of the radio frequency signal based on the AM-AM curve and AM-PM curve that continuously vary with the input power of the radio frequency signal at each discrete frequency point through fitting and cubic spline interpolation algorithm.

[0083] The transfer function construction module 4 is used to construct the linear and nonlinear transfer functions of an RF device based on its forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturation input power curve, noise figure curve, saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, saturation phase distortion curve, and AM-AM and AM-PM curves for each discrete frequency point, which continuously vary with the operating frequency of the RF signal. The linear transfer function characterizes the amplitude and phase linearity of the RF device as it continuously varies with the operating frequency of the RF signal. The nonlinear transfer function characterizes the amplitude and phase nonlinearity of the RF device as it continuously varies with the input power of the RF signal. The nonlinear transfer function has different expressions in the linear, compression, and saturation regions of the RF device.

[0084] The signal transmission calculation module 5 is used to determine whether the RF device is in the linear region, compression region, or saturation region based on the input signal power. When the RF device is in the linear region, the intermediate signal is obtained by using only the linear transfer function for the input signal. When the RF device is in the compression region, the intermediate signal is obtained by using only the nonlinear transfer function for the input signal. When the RF device is in the saturation region, the intermediate signal is obtained by using the fixed parameters for the saturation region.

[0085] The noise transmission calculation module 6 is used to generate complex noise n I +jn Q Depending on whether the RF device is in the linear, compressed, or saturated region, the complex noise is processed in the same way as the input signal of the RF device to obtain complex noise processed by the RF device's transmission characteristics. This processed complex noise is then superimposed onto the intermediate signal to obtain the output signal.

[0086] Compared with existing technologies, the digital twin modeling method for radio frequency devices based on transfer functions proposed in this invention has the following beneficial effects.

[0087] First, in the nonlinear transfer function of the RF device proposed in this invention, a fitting formula including the "difference between input power and input power at the 1dB compression point" is used for the compression region of the RF device. When the RF device is in the compression region, the same fitting formula is applied to both the input signal and noise. The "difference between input power and input power at the 1dB compression point" reflects the current degree of gain compression, and noise processing is based on the "difference between input power and input power at the 1dB compression point," thus realizing the coupled correlation between the noise figure of the RF device and gain compression.

[0088] Second, the present invention uses a cubic spline interpolation algorithm to generate continuous curves from the originally discrete characteristic parameter data. For application scenarios that require rapid switching of operating frequency in a very short time, the value of each characteristic parameter can be obtained in real time without waiting. This avoids the defects of traditional table lookup (dependent on discrete points) or linear interpolation (large error and high delay), and realizes instantaneous response of parameters when the frequency is dynamically switched.

[0089] Third, the input of this invention directly adopts the standard format of the digital prototype link (carrier frequency + IQ signal); in the process, the characteristic parameters of the radio frequency device are directly queried by using the carrier frequency, temperature, current and voltage as indexes, and the instantaneous power is generated by the IQ signal. All calculations are based on the native input data of the digital link; the output of the digital link includes the IQ signal, output frequency, output power and other information required by the digital link, which can be directly connected to the signal link of the digital prototype without any additional format conversion.

[0090] The above are merely preferred embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A digital twin modeling method for radio frequency devices based on transfer function, characterized in that, Includes the following steps; Step S1: Read the forward transmission gain, forward transmission phase, 1dB compression point input power, saturation input power, and noise figure of the RF device at various discrete frequency points. Use the cubic spline interpolation algorithm to interpolate the above RF device characteristic parameters to obtain the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturation input power curve, and noise figure curve of the RF device that change continuously with the operating frequency of the RF signal. Step S2: At each discrete frequency point, input IQ signals of different power to the RF device, and obtain AM-AM data and AM-PM data for each discrete input power point through testing; use cubic spline interpolation algorithm to interpolate the AM-AM data and AM-PM data for each discrete input power point to obtain the AM-AM curve and AM-PM curve that continuously change with the input power of the RF signal at each discrete frequency point. The order of steps S1 and S2 may be either simultaneous or either can be either one before the other. Step S3: Based on the AM-AM curve and AM-PM curve that continuously change with the input power of the RF signal at each discrete frequency point, the saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, and saturation phase distortion curve of the RF device that continuously change with the operating frequency of the RF signal are obtained by fitting and cubic spline interpolation algorithm. Step S4: Based on the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturation input power curve, noise figure curve, saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, saturation phase distortion curve, and AM-AM and AM-PM curves of the RF device continuously varying with the operating frequency of the RF signal, construct the linear transfer function and nonlinear transfer function of the RF device; the linear transfer function characterizes the amplitude linearity and phase linearity of the RF device continuously varying with the operating frequency of the RF signal. The nonlinear transfer function characterizes the amplitude nonlinearity and phase nonlinearity of an RF device as the input power of the RF signal changes continuously; the nonlinear transfer function has different expressions in the linear region, compression region, and saturation region of the RF device; Step S5: Determine whether the RF device is in the linear region, compression region, or saturation region based on the input signal power of the RF device; when the RF device is in the linear region, only the linear transfer function is used to calculate the intermediate signal for the input signal; when the RF device is in the compression region, only the nonlinear transfer function is used to calculate the intermediate signal for the input signal; when the RF device is in the saturation region, the fixed parameters for the saturation region are used to obtain the intermediate signal. Step S6: Generate complex noise; depending on whether the RF device is in the linear region, compression region, or saturation region, process the complex noise in the same way as the input signal of the RF device to obtain complex noise processed by the transmission characteristics of the RF device; superimpose the complex noise processed by the transmission characteristics of the RF device onto the intermediate signal to obtain the output signal.

2. The digital twin modeling method for radio frequency devices based on transfer function according to claim 1, characterized in that, In step S3, P is extracted from the AM-AM curve of a discrete frequency point that continuously varies with the input power of the radio frequency signal. in ≥P insat (f) data, where P in P represents the input signal power. insat (f) represents the saturation input power of the RF device at that discrete frequency point; for P in ≥P insat (f) The saturation gain of the RF device at that discrete frequency point is obtained by fitting the data in segment (f); the saturation gain G of the RF device at each discrete frequency point is obtained in the same way. sat (f) The saturation gain curve of the RF device as the operating frequency of the RF signal changes continuously is obtained by using a cubic spline interpolation algorithm; In step S3, P is extracted from the AM-AM curve of a discrete frequency point that continuously varies with the input power of the radio frequency signal. 1dB (f) < P in <P insat (f) data, where P 1dB (f) represents the 1dB compression point input power of the RF device at this discrete frequency; P 1dB (f) < P in <P insat The formula corresponding to the data in segment (f) , where |S 21 (f) represents the forward transmission gain of the RF device at the discrete frequency point; the compression coefficient of the RF device at the discrete frequency point is obtained by fitting with the least squares method; the compression coefficient α(f) of the RF device at each discrete frequency point is obtained in the same way; the compression coefficient curve of the RF device as the operating frequency of the RF signal changes continuously is obtained by using the cubic spline interpolation algorithm. In step S3, P is extracted from the AM-PM curve of a discrete frequency point that continuously varies with the input power of the radio frequency signal. 1dB (f) < P in <P insat (f) data, P 1dB (f) < P in <P insat The formula corresponding to the data in segment (f) The first and second phase coefficients of the RF device at the discrete frequency point are obtained by fitting with the least squares method; the first phase coefficient β(f) and second phase coefficient γ(f) of the RF device at each discrete frequency point are obtained in the same way; the first phase coefficient curve and second phase coefficient curve of the RF device as the operating frequency of the RF signal changes continuously are obtained by using the cubic spline interpolation algorithm. In step S3, the saturated input power P of the radio frequency device at a certain discrete frequency point is extracted from the AM-PM curve showing the continuous variation of the input power of the radio frequency signal at that discrete frequency point. insat (f) The corresponding phase offset is called the saturation phase distortion value; the saturation phase distortion value of the RF device at each discrete frequency point is obtained in the same way; the saturation phase distortion curve of the RF device when the operating frequency of the RF signal changes continuously is obtained by using the cubic spline interpolation algorithm.

3. The digital twin modeling method for radio frequency devices based on transfer function according to claim 1, characterized in that, In step S4, constructing the linear transfer function of the radio frequency device includes the following steps A1 to A3; Step A1: Convert the forward transmission gain curve of the RF device into a complex function of forward transmission gain |H linear (f)|, converting the forward propagation phase curve of the RF device into a complex function of the forward propagation phase ∠H linear (f); ; where, |S 21 (f)| is the forward transmission gain of an RF device corresponding to a certain operating frequency f, obtained from the forward transmission gain curve; ; where ∠S 21 (f) is the forward transmission phase of the RF device corresponding to a certain operating frequency f, obtained by querying the forward transmission phase curve; Step A2: Complex function of the forward propagation phase ∠H for the RF device linear (f) Differentiate to obtain the continuous group delay function τ(f); ; Step A3: Complex function of forward transmission gain of integrated RF devices |H linear (f)|, Complex function of the forward propagation phase of the radio frequency device ∠H linear (f) The continuous group delay function τ(f) yields the complex form of the linear transfer function H of the RF device. linear (f); Among them, the amplitude term |H linear (f)| Used to apply a frequency-dependent linear gain to the input signal, phase term ∠H linear (f) is used to apply a frequency-dependent linear phase shift to the input signal, and the time delay term -2πfτ(f) is used to correct the time shift of the input signal.

4. The digital twin modeling method for radio frequency devices based on transfer function according to claim 1, characterized in that, In step S4, constructing the nonlinear transfer function of the radio frequency device includes the following steps B1 to B3; Step B1: Based on the forward transmission gain curve, 1dB compression point input power curve, saturation input power curve, saturation gain curve, and compression coefficient curve of the RF device, construct the piecewise gain correction function G. nonlinear (f,P in ); Where f is the operating frequency of the radio frequency signal, and P in It is the input signal power, P 1dB (f) is the 1dB compression point input power obtained from the 1dB compression point input power curve for a certain operating frequency f, P. insat (f) is the saturated input power corresponding to a certain operating frequency f obtained from the saturated input power curve, |S 21 (f)| is the forward transmission gain of the RF device corresponding to a certain operating frequency f, obtained from the forward transmission gain curve; α(f) is the compression coefficient corresponding to a certain operating frequency f, obtained from the compression coefficient curve; G sat (f) is obtained by querying the saturation gain corresponding to a certain operating frequency f from the saturation gain curve; Step B2: Based on the 1dB compression point input power curve, saturation phase distortion curve, first phase coefficient curve, and second phase coefficient curve of the RF device, construct the phase distortion function Δφ(f,P) in ); Where f is the operating frequency of the radio frequency signal, and P in It is the input signal power, P 1dB (f) is the 1dB compression point input power obtained from the 1dB compression point input power curve, β(f) is the first phase coefficient obtained from the first phase coefficient curve, γ(f) is the second phase coefficient obtained from the second phase coefficient curve, and Δφ sat (f) is obtained by querying the saturation phase distortion value corresponding to a certain operating frequency f from the saturation phase distortion curve; The order of steps B1 and B2 may be either simultaneous or either one precedes the other. Step B3: Synthesize the piecewise gain correction function G nonlinear (f,P in ), Phase distortion function Δφ(f,P in The complex form of the nonlinear transfer function H of the radio frequency device is obtained. nonlinear (f,P in ); Where j is the imaginary unit; the real part represents the amplitude nonlinear distortion of the RF device, and the imaginary part represents the phase nonlinear distortion of the RF device; the nonlinear transfer function has three rows, corresponding to the linear region, compression region, and saturation region of the RF device, respectively; in the compression region of the RF device, the phase distortion of the imaginary part is proportional to "the difference between the input power and the input power at the 1dB compression point and the second phase coefficient", allowing the phase distortion to be dynamically adjusted with the input signal power.

5. The digital twin modeling method for radio frequency devices based on transfer function according to claim 1, characterized in that, In step S5, the carrier frequency f and instantaneous power P of the radio frequency device are calculated from the input signal. in 3dB bandwidth B; when the input signal is a single carrier, the carrier frequency of the input signal is extracted by Fast Fourier Transform (FFT); when the input signal is a multi-carrier signal, the subcarrier frequency set of the input signal is extracted by Discrete Fourier Transform (DFT); based on the carrier frequency or subcarrier frequency of the input signal, the 1dB compression point input power corresponding to the carrier frequency or subcarrier frequency is obtained from the 1dB compression point input power curve as the linear region threshold P1, and the saturated input power corresponding to the carrier frequency or subcarrier frequency is obtained from the saturated input power curve as the saturation region threshold P2; Compare input signal power P in Using the linear region threshold P1 and the saturation region threshold P2, determine which operating region the RF device is in; if P in ≤P1 indicates that the RF device is in the linear region; if P1 < P in <P2 indicates that the RF device is in the compressed region; if P in ≥P2 indicates that the radio frequency device is in the saturation region.

6. The digital twin modeling method for radio frequency devices based on transfer function according to claim 5, characterized in that, In step S5, when the RF device is in the linear region, if the input signal is a single-carrier signal, first perform FFT on the input signal to obtain the frequency domain component, then call the linear transfer function, and then perform inverse fast Fourier transform (IFFT) to convert it back to the time domain to obtain the intermediate signal; if the input signal is a multi-carrier signal, first perform FFT on the input signal to obtain the frequency domain component, then call the linear transfer function for each subcarrier, and finally perform IFFT to convert it back to the time domain to obtain the intermediate signal.

7. The digital twin modeling method for radio frequency devices based on transfer function according to claim 5, characterized in that, In step S5, when the radio frequency device is in the compression region, the input signal is first subjected to FFT to obtain the frequency domain component, then the nonlinear transfer function is called, and finally IFFT is performed to convert it back to the time domain to obtain the intermediate signal.

8. The digital twin modeling method for radio frequency devices based on transfer function according to claim 5, characterized in that, In step S5, when the radio frequency device is in the saturation region, the fixed parameters of the saturation region are called to output the intermediate signal.

9. The digital twin modeling method for radio frequency devices based on transfer function according to claim 1, characterized in that, Step S6 specifically includes the following steps S61 to S65; Step S61: Calculate the noise power spectral density function N0 based on the carrier frequency f and 3dB bandwidth B of the input signal of the RF device; convert the noise power spectral density function N0 into a linear power P. n ; Step S62: Distribute the complex noise power to the I and Q channels, and calculate the power P of each channel. nI and P nQ Variance of each path ; Step S63: Generate two independent zero-mean values ​​with powers P and P respectively. nI and P nQ , and variance are respectively and A Gaussian white noise sequence, i.e., I-channel in-phase noise n I Orthogonal noise of Q-path n Q The combination forms time-domain complex noise n I +jn Q ; Step S64: If the input signal of the RF device is a single-carrier signal and the RF device is in the linear region, only the linear transfer function is called for the time-domain complex noise to obtain the complex noise processed by the transmission characteristics of the RF device; If the input signal of the RF device is a single-carrier signal and the RF device is in the compression region, only the nonlinear transfer function is called for the time-domain complex noise to obtain the complex noise processed by the transmission characteristics of the RF device; If the input signal of the RF device is a single-carrier signal and the RF device is in the saturation region, using fixed parameters in the saturation region will produce complex noise processed by the transmission characteristics of the RF device. If the input signal of the RF device is a multi-carrier signal and the RF device is in the linear region, the time-domain complex noise is first converted into frequency-domain noise through FFT. Only the linear transfer function is called on the frequency-domain noise. Then, the processed frequency-domain noise is converted back to the time domain through IFFT. This is the complex noise processed by the transmission characteristics of the RF device. If the input signal of the RF device is a multi-carrier signal and the RF device is in the compression region, the time-domain complex noise is first converted into frequency-domain noise through FFT. Only the nonlinear transfer function is called for the frequency-domain noise. Then, the processed frequency-domain noise is converted back to the time domain through IFFT. This is the complex noise processed by the transmission characteristics of the RF device. If the input signal of the RF device is a multi-carrier signal and the RF device is in the saturation region, the time-domain complex noise is first converted into frequency-domain noise through FFT. The frequency-domain noise is output with fixed parameters in the saturation region. Then, the output frequency-domain noise is converted back to the time domain through IFFT. This is the complex noise processed by the transmission characteristics of the RF device. Step S65: The complex noise processed by the transmission characteristics of the radio frequency device is superimposed on the intermediate signal to obtain the output signal. The output signal is encapsulated in the format of the standardized signal link of the digital prototype.

10. A digital twin modeling system for radio frequency devices based on transfer function, characterized in that, It includes a continuous frequency curve generation module 1, a continuous power curve generation module, a continuous frequency curve generation module 2, a transfer function construction module, a signal transfer calculation module, and a noise transfer calculation module; The continuous frequency curve generation module is used to read the forward transmission gain, forward transmission phase, 1dB compression point input power, saturation input power, and noise figure of the RF device at various discrete frequency points. The cubic spline interpolation algorithm is used to interpolate the above RF device characteristic parameters to obtain the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturation input power curve, and noise figure curve of the RF device that change continuously with the operating frequency of the RF signal. The continuous power curve generation module is used to input IQ signals of different power to the radio frequency device at each discrete frequency point. The AM-AM data and AM-PM data of each discrete input power point are obtained through testing. The cubic spline interpolation algorithm is used to interpolate the AM-AM data and AM-PM data of each discrete input power point to obtain the AM-AM curve and AM-PM curve that change continuously with the input power of the radio frequency signal at each discrete frequency point. The continuous frequency curve generation module 2 is used to obtain the saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, and saturation phase distortion curve of the radio frequency device that continuously varies with the operating frequency of the radio frequency signal based on the AM-AM curve and AM-PM curve that continuously vary with the input power of the radio frequency signal at each discrete frequency point through fitting and cubic spline interpolation algorithm. The transfer function construction module is used to construct the linear and nonlinear transfer functions of an RF device based on the forward transmission gain curve, forward transmission phase curve, 1dB compression point input power curve, saturation input power curve, noise figure curve, saturation gain curve, compression coefficient curve, first phase coefficient curve, second phase coefficient curve, saturation phase distortion curve, and AM-AM and AM-PM curves that continuously change with the input power of the RF signal at each discrete frequency point. The linear transfer function characterizes the amplitude linearity and phase linearity of the RF device as it continuously changes with the operating frequency of the RF signal. The nonlinear transfer function characterizes the amplitude nonlinearity and phase nonlinearity of an RF device as the input power of the RF signal changes continuously; the nonlinear transfer function has different expressions in the linear region, compression region, and saturation region of the RF device; The signal transmission calculation module is used to determine whether the RF device is in the linear region, compression region, or saturation region based on the input signal power of the RF device. When the RF device is in the linear region, it uses only the linear transfer function to calculate the intermediate signal. When the RF device is in the compression region, it uses only the nonlinear transfer function to calculate the intermediate signal. When the RF device is in the saturation region, it uses the fixed parameters of the saturation region to obtain the intermediate signal. The noise transmission calculation module is used to generate complex noise; depending on whether the RF device is in the linear region, compression region, or saturation region, the complex noise is processed in the same way as the input signal of the RF device to obtain complex noise processed by the transmission characteristics of the RF device; the complex noise processed by the transmission characteristics of the RF device is superimposed on the intermediate signal to obtain the output signal.

Images